Qwen3.8-Flash-Next Uncensored for Basalt

Basalt model files of orcarouter's abliterated (refusal-direction-removed) Qwen3.8-Flash-Next. They are drop-in replacements for the official Qwen3.8-FN-NVFP4-DQ6_K and Qwen3.8-FN-NVFP4-DQ8 files of jesdga/Qwen3.8-Flash-Next-Basalt: same format (.basalt version 2, expert layout v6), same tensor types and sizes, and they run on the unmodified engine.

Safety alignment has been removed. The model complies with requests the original refuses. It is meant for research (interpretability, refusal mechanisms, red-teaming); you are responsible for how it is used and for adding your own safeguards before anything reaches end users. See orcarouter's model card for how the edit was made.

What each part comes from

Sources: orcarouter/Qwen3.8-Flash-Next-Uncensored (BF16, revision da29941a), orcarouter/Qwen3.8-Flash-Next-Uncensored-NVFP4 (revision 9c6f1345), and the official Qwen3.8-FN-NVFP4-DQ6_K.basalt (revision 21b83575).

The drafter is the official self-distilled one. Speculative decoding does not change output, since the main model verifies every drafted token.

Every tensor the abliteration did not edit is byte-identical to the official DQ6_K file (attention q/k/v, GDN projections and parameters, shared-expert gate/up, head, router, hyper-connections and their hc8 copies, indexer, PLE key, norms). The ones that differ are exactly the edited residual writers (attn_output, ssm_out, ffn_down_shexp, ple_value, token_embd), the experts and the n-gram table.

Chat template

The files carry froggeric's fixed Qwen template instead of the official Qwen3.8 one. The server renders it; the engine is unchanged. What differs from the official template:

• Default reasoning effort is medium, not xhigh. A request that sets no effort thinks less than with the official files. Send reasoning_effort: "high" (the server maps it to xhigh) for the official default.
• Tool-call arguments sent as a JSON string render instead of failing, and the tool-use instructions in the system prompt are reworded (reasoning inside <think>, the <tool_call> right after it).
• Inline <|think_low|>, <|think_medium|>, <|think_xhigh|>, <|think_off|> tags in a message steer the effort.
• Consecutive leading system and developer messages are merged into one system turn.

The quality numbers below are unaffected: the gate scores raw text and does not use a template. The server keys its extracted tokenizer cache by the meta section's hash, so a cache from an earlier copy of these files is not reused.

Quality

Measured with basalt-bench --quality (tools/ppl_gate.py): 100 chunks of wikitext-2 (512 tokens each), the second half of each chunk scored through the decode path, against a BF16 reference of the original Qwen/Qwen3.8-Flash-Next (revision de4b8e4d), made the way the official card's is: convert_hf_to_gguf.py --outtype bf16 and llama-perplexity --kl-divergence-base at llama.cpp 436f6f89e, -c 512 -b 8192 -ub 8192, experts on the CPU (reference PPL 3.2778 on the scored rows). T 3, int8 KV, resident experts, one RTX 5090 at its default 575 W power limit (the limit could not be lowered on the rented machine). All three rows were measured on the same machine against the same reference.

The reference is the unedited model, so these numbers include the abliteration's intended change to the output distribution as well as quantization. The quantization itself is the official recipe (every unedited tensor is byte-identical to the official file; edited ones go through the same single rounding), and DQ8 is 0.0071 KLD closer than DQ6_K, as with the official pair (0.0063 on the official card). The re-measured official row agrees with the official card (KLD 0.0624, top-1 91.65%, measured on different hardware).

Run

As for the official files, with model pointing at one of these:

hf download klee100/Qwen3.8-Flash-Next-Uncensored-Basalt Qwen3.8-FN-Uncensored-NVFP4-DQ6_K.basalt --local-dir models
bin/basalt serve --config basalt_config.json

The expert profile in the engine's data/profiles was learned on the original weights. It still applies (same architecture and expert count), but which experts are hottest may differ slightly.

Build

python3 tools/convert/convert.py --sections-from Qwen3.8-FN-NVFP4-DQ6_K.basalt \
    --checkpoint Qwen3.8-Flash-Next-Uncensored --experts-checkpoint Qwen3.8-Flash-Next-Uncensored-NVFP4 \
    --out Qwen3.8-FN-Uncensored-NVFP4-DQ6_K.basalt --name "Qwen3.8-Flash-Next Uncensored NVFP4"
# add --dense-q8-0 for the DQ8 file

--experts-checkpoint (compressed-tensors experts) and the BF16 n-gram table reader are converter additions not yet in the upstream repository.

License

Qwen Community License 1.0 (LICENSE), the license of Qwen3.8-Flash-Next and of the LICENSE file in orcarouter's repository. orcarouter's model card front matter says Apache-2.0, and that value is carried in the files' GGUF metadata (general.license); the license text embedded in each file is the Qwen Community License.